Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a – 3 Theories of Learning 3.1 Overview of Learning Theories 3.2 Behaviorism 3.3 Cognitivism 3.4 Constructivism 3.5 Learning as Social Process Literature: B.R. Hergenhahn, M. H. Olson: An Introduction to Theories of Learning, 6th edition, Prentice-Hall 2001, Revised edition 2012 1
34
Embed
3 Theories of Learning - LMU Medieninformatik · 3 Theories of Learning ... Classical Conditioning: Ivan Petrovich Pavlov Ivan Pavlov (1849 ... – Signal learning (similar to Pavlov‘s
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
3 Theories of Learning3.1 Overview of Learning Theories 3.2 Behaviorism 3.3 Cognitivism 3.4 Constructivism 3.5 Learning as Social Process
Literature:B.R. Hergenhahn, M. H. Olson: An Introduction to Theories of Learning, 6th edition, Prentice-Hall 2001, Revised edition 2012
1
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Learning is a Difficult Topic: Imprinting• Konrad Lorenz (1952):
– Chicklets of geese (Greylag geese) bond with the first moving stimulus they perceive – even if this is not a goose but a human being
– Change of behavior typical for species• Did the chicklets "learn"?
– Is this "unlearned behavior"?– No "reinforced practice"– Definitely "experience"-based
2
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Lerner
Input Output
Feedback
Lerner
Input Output
Feedback
Lerner
Input Output
Theories of Learning: OverviewBehaviorism Cognitivism Construktivism
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
3 Theories of Learning3.1 Overview of Learning Theories 3.2 Behaviorism 3.3 Cognitivism 3.4 Constructivism 3.5 Learning as Social Process
Literature:B.R. Hergenhahn, M. H. Olson: An Introduction to Theories of Learning, 6th edition, Prentice-Hall 2001, Revised edition 2012http://www.psywww.com/intropsych/ch05_conditioning/
4
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Classical Conditioning: Ivan Petrovich Pavlov
Ivan Pavlov (1849 – 1936)
5
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Conditioning• Unconditioned stimulus
(US)– e.g. meat
• Conditioned stimulus (CS)
– e.g. sound• Unconditioned reaction
(UR)– e.g. saliva
• Conditioned reaction (CR)
– e.g. saliva• CR and UR
– same quality– CR lower magnitude
than UR
• Unconditioned behavior:US → UR
• Training:US + CS → UR
• Conditioned behavior:CS → CR
Bild: http://www.acs.appstate.edu
6
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Examples of Classical Conditioning
• Please think about examples in daily life where classical (Pavlov-like) conditioning takes place for humans!
7
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Paradigm of Behaviorism
• Elaborate the laws of the relationship between stimulus and response• Learning in behaviorism:
– To condition the responses of the learning subjects to certain stimuli• Not only Pavlov-like classical conditioning!
8
Human being=
Black BoxStimulus Response
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Edward Lee Thorndike: Connectionism• Connection =
neural connection between stimulus and response• Assumption: All mammals learn in the same manner
– Experiments with animals (e.g. cats and monkeys)• Implicit assumption:
– There is no reasoning involved in learning
Edward Thorndike (1874 – 1949)
Thorndike puzzle boxes for cats
Before the main behavioristic movement!
9
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Thorndike: Law of Effect
• Response followed by a reward → strength of connection is increased
• Response followed by a punishment→ strength of connection is decreased
• Revised law of effect (1930)– rewards work for reinforcement of connections– punishments do not influence the strength of a connection
10
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
– Rejects to use terms like “drive”, “motivation”• Behavior
– Respondent behavior» Elicited by known stimulus» Controlled by its causes» “Type S” conditioning (Pavlov-like)
– Operant behavior» Not elicited by known stimulus » Just "emitted" by organism, seems to appear spontaneously» Probability of certain behavior is modified according to consequences» “Type R” conditioning (operant conditioning)
B.F. Skinner (1904 – 1990)
11
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Principles of Operant Conditioning• A response followed by a reinforcing stimulus tends to be
repeated.• “The only defining characteristic of a reinforcing stimulus is that it
reinforces.”
To modify behavior:• Find something that is reinforcing• Wait until desired behavior appears• Immediately reinforce!
responsereinforcer
leads to
responseresponseresponse…
12
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
The Skinner Box• Grid floor (can be used for electric shocks)• Light• Lever• Food cup (reinforcement)
13
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
– “A pigeon is brought to a stable state of hunger by reducing it to 75 percent of its weight when well fed. It is put into an experimental cage for a few minutes each day. A food hopper attached to the cage may be swung into place so that the pigeon can eat from it. A solenoid and a timing relay hold the hopper in place for five sec. at each reinforcement.” (B.F. Skinner)
• Automated collection of data:
14
'SUPERSTITION' IN THE PIGEON B. F. Skinner, Indiana UniversityFirst published in Journal of Experimental Psychology, 38, 168-172 (1947)
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Discriminative Operant Conditioning
• Combination of cause- and effect-based conditioning• Example:
Light switched on → animal presses lever → food is dispenseddiscriminative → operant response → reinforcing stimulus
stimulus (SD) → (R) → (SR)
Association of interest
15
Is this the same as classical (Pavlov-style) conditioning?
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Chaining
• Chaining: Reinforcing stimulus of one response acts as discriminative stimulus for another response
• Backward chaining: Adding a new stimulus to conditioned behavior • Example:
in test chamber → orient toward lever → light switched on → presses lever → …
(SD) → (R) → (SR) (SD) → (R) → …
16
Could we use this technique to learn animals to do tricks?
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Shaping of Behavior• Over time, step-wise
reinforcement of behaviors shapes behavior.
• Example: – How can we make pigeons to
turn clockwise (completely)?– Please describe how you
would approach this task!• Is this learning? teaching?
17
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Watch B. F. Skinner with a Conditioned Pigeon
18
www.youtube.com/watch?v=TtfQlkGwE2U
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Superstitious Behavior• What happens if the reinforcement appears randomly,
independent of what the subject animal is doing?– E.g. food is dispensed at random times
• Can you give examples of superstitious behavior of this kind in humans?
20
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Example: Multi-Armed Bandit
21
Wright, John C.: »Consistency and Compexity of Response Sequences as a Function of Schedules of Noncontingent Reward« Journal of Experimental Psychology 63 : 601-9, 1962.
Read it up in:Paul Watzlawick. How Real is Real? Confusion, Disinformation, Communication. New York:Vintage, 1977(Dt.: Wie wirklich ist die Wirklichkeit?)See also http://www.alex-sk.de/D_Wright.html
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
• Estes (1944)– Punishment turned out as effective only in a very short time range– In the long run no more effective than „extinction“
• Skinner: Short term effect of punishment reinforces the punisher.
22
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Criticism of Behaviorism• (to be completed in classroom)• Please list criticisms on the behaviorism approach to learning!
23
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
3 Theories of Learning3.1 Overview of Learning Theories 3.2 Behaviorism 3.3 Cognitivism 3.4 Constructivism 3.5 Learning as Social Process
Literatur:B.R. Hergenhahn, M. H. Olson: An Introduction to Theories of Learning, 6th edition, Prentice-Hall 2001, Revised edition 2012
24
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
The Gestaltists: Focus on the Human Being• Kurt Lewin (1890 – 1947):
– Field theory of human motivation– All psychological facts a human experiences make up a person‘s life space.– The totality of these events determines behavior at any given time.– A person exists in a continually changing field of influences, and a change in
one of them affects all the others.– Active role of the brain: acts on sensory information
• Max Wertheimer (1880 – 1943)– Gestalt laws for human perception– Rules for subjective interpretation of information by humans
25
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Gestalt Law: Law of Familiarity• Human perception groups elements which give a known meaning.• Several different interpretations of the same information are possible.
27
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Wolfgang Köhler: Problem Solving in Apes• Experiments with chimpanzees
– Usage of tools, combinations, …• Chickens would not be able to do that!
– „Insightful learning“
Wolfgang Köhler (1887 – 1967)
28
www.youtube.com/watch? v=FwDhYUlbxiQ
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –
Wolfgang Köhler: TranspositionAn experiment beyond stimulus-response connections:
Stimuli during Preliminary Training:Animal is fed only on light grey surface
Stimuli during Transposition Test:Which surface is preferred, light or dark grey?
Transposition: Principle learned in one problem,applied to another problem
29
Ludwig-Maximilians-Universität München Prof. Hußmann Multimediale Lehr- und Lernumgebungen, SS 2015 – 3a –